An accurate extended Kalman filter scheme for the reduction of the flux angle tracking error in sensorless induction motor drives

The paper deals with extended Kalman filters for the estimation of rotor flux and angular velocity in induction motors. It investigates the impact of discretization errors on the performance of extended Kalman filters in sensorless induction motor drives. It shows that first order Euler approximation and second order approximations may lead to insufficient flux angular accuracy when the sampling frequency is low or the stator frequency is high. It proposes to use an extended Kalman filter that estimates and compensates the discretization error. It shows that this solution allows to substantially reduce the flux angular error below what can be obtained with first and second order approximations. The scheme is verified experimentally.